Wyniki wyszukiwania - BazTech

1

Recurrent neural network optimization for wind turbine condition prognosis

Kerboua Adlen, Kelaiaia Ridha

Diagnostyka

|

2022

|

Vol. 23, No. 3

art. no. 2022301

EN

This research focuses on employing Recurrent Neural Networks (RNN) to prognosis a wind turbine operation’s health from collected vibration time series data, by using several memory cell variations, including Long Short Time Memory (LSTM), Bilateral LSTM (BiLSTM), and Gated Recurrent Unit (GRU), which are integrated into various architectures. We tune the training hyperparameters as well as the adapted depth and recurrent cell number of the proposed networks to obtain the most accurate predictions. Tuning those parameters is a hard task and depends widely on the experience of the designer. This can be resolved by integrating the training process in a Bayesian optimization loop where the loss is considered as the objective function to minimize. The obtained results show the effectiveness of the proposed method, which generates more accurate recurrent models with a more accurate prognosis of the operating state of the wind turbine than those generated using trivial training parameters.

2

Deep reinforcement learning overview of the state of the art

Fenjiro Y., Benbrahim H.

Journal of Automation Mobile Robotics and Intelligent Systems

|

2018

|

Vol. 12, No. 3

20--39

EN

Artificial intelligence has made big steps forward with reinforcement learning (RL) in the last century, and with the advent of deep learning (DL) in the 90s, especially, the breakthrough of convolutional networks in computer vision field. The adoption of DL neural networks in RL, in the first decade of the 21 century, led to an end-toend framework allowing a great advance in human-level agents and autonomous systems, called deep reinforcement learning (DRL). In this paper, we will go through the development Timeline of RL and DL technologies, describing the main improvements made in both fields. Then, we will dive into DRL and have an overview of the state-ofthe- art of this new and promising field, by browsing a set of algorithms (Value optimization, Policy optimization and Actor-Critic), then, giving an outline of current challenges and real-world applications, along with the hardware and frameworks used. In the end, we will discuss some potential research directions in the field of deep RL, for which we have great expectations that will lead to a real human level of intelligence.

3

Intelligent LQI-based wireless sensor network applied to ZigBee positioning system

Mon Y. J., Lin C. M.

Przegląd Elektrotechniczny

|

2013

|

R. 89, nr 5

241-244

EN

In this paper, a link quality indicator (LQI) based wireless sensor network (WSN) constructed by a recurrent fuzzy neural network (RFNN) is developed as a ZigBee Positioning System (ZPS) to monitor and realize the tag of 802.15.4/ZigBee locations. First, the performance of LQI is demonstrated, then it is applied to develop a ZPS which is used to verify the performance of indoor location identification. Finally, an RFNN is used to combine with the ZPS to develop a location system, and it can be applied for children’s position monitoring. The experimental results demonstrate good positioning performance has been achieved by the proposed location system.

PL

W artykule opisano sieć czujników bezprzewodowych zbudowaną za pomocą sieci neuronowej RFNN, na bazie metody LQI. Ma ona zastosowanie w protokole 802.15.4/ZigBee jako blok lokalizacji (ZigBee Positioning System). Omówione zostało zastosowanie LQI, który został wykorzystany w projektowaniu ZPS. Na koniec wykorzystano RFNN oraz ZPS w budowie systemu lokalizacji. Badania eksperymentalne potwierdziły skuteczność działania proponowanego systemu.

4

Evolving adaptive recurrent networks with Developmental Symbolic Encoding

Suchorzewski M.

Metody Informatyki Stosowanej

|

2010

|

nr 1 (22)

175-187

EN

In this paper we are concerned with evolutionary synthesis of recurrent networks capable of learning in the environment. First, we define the model of network we aim to evolve, which is weightless recurrent network of basic arithmetic nodes. Next, we propose a developmental genetic representation for the network, along with some genetic operators for it. The representation bears some important characteristics such as closure and completeness. Most notably, however, it features modularity and scalability, which we demonstrate on a parity problem. Finally, we evolve the network capable of successful learning in some narrow problem domain. The result shows, that for a given problem domain, evolutionary approach may produce networks performing better than generic neural networks.